Ian R. White

Does the variation in the socioeconomic characteristics of an area affect mortality

Our research in England has shown that the more deprived an area the greater its incidence of premature mortality.1 Wilkinson has argued that in the developed world income distribution is a more important predictor of life expectancy between countries than simply mean income.2 We aimed to determine whether the risk of mortality in a geographical area was related to the degree of socioeconomic variation within that area as well as the average level of deprivation. For each of the 8464 wards in England we obtained the Townsend deprivation index from the 1981 census1 and directly standardised all cause mortality for 1981-5. Mortality under the age of 65 was used as an indicator of premature mortality. Male and female mortality rates were averaged for each ward. Twenty four wards were …

Dietary Fiber and Colorectal Cancer Risk: A Nested Case–Control Study Using Food Diaries

BACKGROUND Results of epidemiological studies of dietary fiber and colorectal cancer risk have not been consistent, possibly because of attenuation of associations due to measurement error in dietary exposure ascertainment. METHODS To examine the association between dietary fiber intake and colorectal cancer risk, we conducted a prospective case-control study nested within seven UK cohort studies, which included 579 case patients who developed incident colorectal cancer and 1996 matched control subjects. We used standardized dietary data obtained from 4- to 7-day food diaries that were completed by all participants to calculate the odds ratios for colorectal, colon, and rectal cancers with the use of conditional logistic regression models that adjusted for relevant covariates. We also calculated odds ratios for colorectal cancer by using dietary data obtained from food-frequency questionnaires that were completed by most participants. All statistical tests were two-sided. RESULTS Intakes of absolute fiber and of fiber intake density, ascertained by food diaries, were statistically significantly inversely associated with the risks of colorectal and colon cancers in both age-adjusted models and multivariable models that adjusted for age; anthropomorphic and socioeconomic factors; and dietary intakes of folate, alcohol, and energy. For example, the multivariable-adjusted odds ratio of colorectal cancer for highest vs the lowest quintile of fiber intake density was 0.66 (95% confidence interval = 0.45 to 0.96). However, no statistically significant association was observed when the same analysis was conducted using dietary data obtained by food-frequency questionnaire (multivariable odds ratio = 0.88, 95% confidence interval = 0.57 to 1.36). CONCLUSIONS Intake of dietary fiber is inversely associated with colorectal cancer risk. Methodological differences (ie, study design, dietary assessment instruments, definition of fiber) may account for the lack of convincing evidence for the inverse association between fiber intake and colorectal cancer risk in some previous studies.

Green Tea Consumption and Serum Lipids and Lipoproteins in a Population of Healthy Workers in Japan

PURPOSE To examine the relation between green tea consumption and serum lipids and lipoproteins. METHODS The subjects were 13,916 workers (8476 men and 5440 women) aged 40-69 years at over 1000 workplaces in Nagano prefecture, central Japan. They underwent health screening offered by a single medical institute between April 1995 and March 1996 and did not have morbid conditions affecting serum cholesterol levels. Serum concentrations of total cholesterol, high-density lipoprotein (HDL) cholesterol and triglycerides were measured at the screening. The consumption of green tea and other life-style characteristics were ascertained by a questionnaire. The data were analyzed with multivariate linear model. RESULTS Daily consumption of green tea was reported by 86.7% of subjects. Green tea consumption was, statistically, significantly associated with lower levels of serum total cholesterol in both men and women while its associations with serum triglycerides and HDL cholesterol were not statistically significant. The inverse association of serum total cholesterol with green tea consumption appeared to level off at the consumption of more than 10 cups/day. Excluding the outlying subjects drinking more than 10 cups/day (0.4%), the regression analysis adjusting for age, body mass index, ethanol intake, smoking habit, coffee intake, and type of work showed that daily consumption of one cup of green tea was associated with a reduction in serum total cholesterol by 0.015 mmol/L (95% confidence interval 0.006 to 0.024, p < 0.001) in men and 0.015 mmol/L (0.004 to 0.025, p < 0.01) in women. After additional adjustment for selected dietary factors, the inverse association remained statistically significant; one cup of green tea per day was associated with a reduction in serum total cholesterol by 0.010 mmol/L (0.001 to 0.019, p = 0.03) in men and 0.012 mmol/L (0.001 to 0.022, p = 0.03) in women. CONCLUSION Consumption of green tea was associated with lower serum concentration of total cholesterol in Japanese healthy workers age 40-69 years; however, green tea consumption was unrelated to serum HDL-cholesterol and triglycerides.

Gender and employment grade differences in blood cholesterol, apolipoproteins and haemostatic factors in the Whitehall II study

In the first Whitehall study, plasma cholesterol was a strong predictor of coronary heart disease (CHD) but it showed a positive association with grade of employment: the higher the grade the higher the level. Because it could not explain the higher rate of CHD in lower employment grades, further investigation of biochemical CHD risk factors has been conducted with data from the baseline examination of the Whitehall II cohort in 1985-88. These data also allow investigation of gender differences and the effect of menopause. Serum cholesterol (6860 men and 3374 women) and apolipoproteins A-I and B (apo AI and apo B) were measured in those aged 35-55 working in the London offices of twenty Civil Service departments. Plasma fibrinogen and factor VII were determined in 45-55 year olds. The apo B/apo AI ratio (95% confidence interval) after age adjustment is lowest in premenopausal women: 0.557 (0.549-0.565), intermediate in postmenopausal women: 0.601 (0.589-0.613) and highest in men: 0.703 (0.698-0.709). After age adjustment fibrinogen is higher in postmenopausal (2.90 (2.85-2.95) g/l) than in premenopausal women (2.78 (2.71-2.84) g/l), who have higher levels than men (2.64 (2.62-2.67) g/l). A positive association with employment grade is seen for apo AI and a negative association is seen for fibrinogen, apo B (women only) and the apo B/apo AI ratio, after age adjustment. These patterns are consistent with the higher rates of CHD in lower grades. Cholesterol and factor VII show no gradient with our sensitive measure of social position. After adjusting for the effects of smoking rates, alcohol consumption, exercise and dietary pattern, as well as age, ethnicity, body mass index and report of symptoms, the regression coefficient for apo AI on employment grade is reduced by 43% in men and 70% in women. Corresponding reductions for fibrinogen are 53% and 65%. These attenuations suggest that a considerable part of the social gradients in apo AI and fibrinogen are explained by variations in health related behaviours. The remaining gradients may represent effects independent of these behaviours.

The Level of Alcohol Consumption at Which All-Cause Mortality Is Least

Moderate consumers of alcohol have lower mortality than either nondrinkers or heavy drinkers. This systematic review aimed to quantify the level of alcohol consumption (termed the nadir) at which the lowest mortality occurs. Twenty cohort studies reported analyses of all-cause mortality for at least three categories of alcohol consumption, giving a total of 60,224 deaths among men and 74,824 deaths among women. The nadir in each study was estimated for men and women separately in units per week, where 1 unit is 9 g of alcohol. The estimated nadirs varied substantially between countries. Combined nadirs were estimated for U.S. men (overall nadir 7.7 units per week, 95% confidence interval [CI] 6.4-9.1), U.K. men (12.9 units per week, 95% CI 10.8-15.1), and U.S. women (2.9 units per week, 95% CI 2.0-4.0). The nadirs were not found to be increased in studies of older persons and apply for ages 50 to 80 years.

Randomization-based methods for correcting for treatment changes: examples from the Concorde trial.

We develop analysis methods for clinical trials with time-to-event outcomes which correct for treatment changes during follow-up, yet are based on comparisons of randomized groups and not of selected groups. A causal model relating observed event times to event times that would have been observed under other treatment scenarios is fitted using the semi-parametric approach of Robins and Tsiatis (avoiding assumptions about the relationship between treatment changes and prognosis). The methods are applied to the Concorde trial of immediate versus deferred zidovudine, to investigate how the results would have differed if no participant randomized to deferred zidovudine had started treatment before reaching ARC or AIDS. We consider issues relating to model choice, non-constant treatment effects and censoring.

Clinical trials comparing two treatment policies: which aspects of the treatment policies make a difference?

We discuss pragmatic clinical trials with survival endpoints in which subjects commonly change treatment during follow-up. Suppose that an intention-to-treat (ITT) analysis shows a significant difference between the randomized groups. We may want to ask questions about the reason for such a difference in outcome between randomized groups: for example, was the difference due to different policies for change to a third more beneficial regime? We address such questions using the semi-parametric accelerated life models of Robins, which exploit the randomization assumption fully and avoid direct comparisons of possibly differently selected subgroups. No assumption is made about the relationship of treatment actually prescribed to prognosis. A sensitivity analysis, using a range of plausible values for the causal effect of a covariate, estimates the contrasts between randomized groups that would have been observed if the covariate had universally been 0. The main technical problem is in dealing with censoring, for the method requires different degrees of recensoring for different values of the causal effect, and this can lead to estimates of low precision. The methods are applied to a randomized comparison of two anti-hypertensive treatments in which approximately half the subjects changed treatment during follow-up. Various time-dependent covariates, representing patterns of side-effects and treatments, are used in the model. We find that the observed difference in cardiovascular deaths between the randomized groups cannot be explained in this way by their different covariate patterns.

Rose questionnaire angina in younger men and women: gender differences in the relationship to cardiovascular risk factors and other reported symptoms.

Cross-sectional data from the Whitehall II study baseline were used to identify factors that may lead to the high levels of Rose angina reporting in women. 134 (4.0%) of 3350 women and 164 (2.4%) of 6830 men reported angina (P<0.001). Women with Rose angina had a poorer cardiovascular risk profile (degree of obesity, serum cholesterol and apolipoprotein B, blood pressure) and more electrocardiogram abnormalities (ST and T changes) than women without angina, but the associations were generally weaker than in men. Women who reported many other physical symptoms had a high prevalence of Rose angina (9.7%). Adjustment for symptom reporting reduced the age-adjusted gender difference to odds ratio (OR) = 0.93 (95% confidence interval [CI]: 0.56-1.56) for subjects with no symptoms, and to OR = 1.42 (95% CI = 1.05-1.90) for subjects at the upper quartile of symptom score. Among women a high level of general symptom reporting was associated with General Health Questionnaire (GHQ) minor psychiatric morbidity (51.9% prevalence), but GHQ caseness does not appear to be a predictor of Rose angina (OR 1.22 [0.67-2.21]) in this group. Coronary artery disease risk is raised in women with Rose angina, and this remains true in groups with high levels of general symptom reporting.

Assessing Risk Prediction Models Using Individual Participant Data From Multiple Studies

Individual participant time-to-event data from multiple prospective epidemiologic studies enable detailed investigation into the predictive ability of risk models. Here we address the challenges in appropriately combining such information across studies. Methods are exemplified by analyses of log C-reactive protein and conventional risk factors for coronary heart disease in the Emerging Risk Factors Collaboration, a collation of individual data from multiple prospective studies with an average follow-up duration of 9.8 years (dates varied). We derive risk prediction models using Cox proportional hazards regression analysis stratified by study and obtain estimates of risk discrimination, Harrells concordance index, and Roystons discrimination measure within each study; we then combine the estimates across studies using a weighted meta-analysis. Various weighting approaches are compared and lead us to recommend using the number of events in each study. We also discuss the calculation of measures of reclassification for multiple studies. We further show that comparison of differences in predictive ability across subgroups should be based only on within-study information and that combining measures of risk discrimination from case-control studies and prospective studies is problematic. The concordance index and discrimination measure gave qualitatively similar results throughout. While the concordance index was very heterogeneous between studies, principally because of differing age ranges, the increments in the concordance index from adding log C-reactive protein to conventional risk factors were more homogeneous.

The Development of a Simulation Model of Primary Prevention Strategies for Coronary Heart Disease

This paper describes the present state of development of a discrete-event micro-simulation model for coronary heart disease prevention. The model is intended to support health policy makers in assessing the impacts on health care resources of different primary prevention strategies. For each person, a set of times to disease events, conditional on the individuals risk factor profile, is sampled from a set of probability distributions that are derived from a new analysis of the Framingham cohort study on coronary heart disease. Methods used to model changes in behavioural and physiological risk factors are discussed and a description of the simulation logic is given. The model incorporates POST (Patient Oriented Simulation Technique) simulation routines.